Abstract
Typically, one describes the planning of human arm movements in terms of a vector from an initial position (e.g., of the fingertip) to a target. In addition, movement plans might make use of egocentric coordinates (e.g., the joint angles required to land on the target). If movement errors used to correct subsequent movements are coded in vector terms, then performance in a series of reaches will be best when movements are blocked by initial-position-to-target vector. If, in addition, humans make use of egocentric coordinates for coding errors, performance should be better when movements are blocked by egocentric target location, independent of the initial position. We compare these two hypotheses in a reaching task.
Methods: Subjects made reaches on a tabletop. Initial position, current fingertip location and target position were shown on a fronto-parallel display. Reaches included all combinations of 6 reach vectors and 6 egocentric targets (36 distinct reaches) within two contexts. In the ‘egocentric’ context, reaches were blocked according to target location independent of initial position; all reaches to a given target were completed before reaches to any other target were made. In the ‘vector’ context, reaches were grouped according to reach direction without respect to target identity. In each context, 20 repetitions of each reach were made (randomized within blocks). The order of the two contexts and of blocks within context were randomized across subjects. Target size was tuned by subject to equate task difficulty.
Results: Performance, measured in terms of target hit rate, was superior in the egocentric context. This supports the hypothesis that when the reach context affords its use, error information obtained from egocentric target coding is combined with vector coding to improve human movement planning.
Acknowledgements: NIH EY08266